/KnowPrompt

Primary LanguagePython

File Structure

## label_data_with_gpt.py   
    --data_id  
    --prompt_id
    --llm_id 

## LabeledDaraset/
    LabeledDaraset/{LLM_id}-{task_id}-{prompt_id}/train
    LabeledDaraset/{LLM_id}-{task_id}-{prompt_id}/test

## preprocess_dataset.py    preprocess spider dataset

## prompt_template.py       manul defined prompt

## train_local_model.py
    --small_dnn_id
    --labeled_data_id

## utils.py              common used function

Step

1. get the ground truth dataset for each task.  
    spider dataset needs specifical preprocess

2. query gpt and save to ```LabeledDaraset```

3. train local model using the dataset from ```LabeledDaraset```.